Differentiable Safe Controller Design Through Control Barrier Functions

• Published in: IEEE control systems letters Vol. 7; pp. 1207 - 1212
• Main Authors: Yang, Shuo, Chen, Shaoru, Preciado, Victor M., Mangharam, Rahul
• Format: Journal Article
• Language: English
• Published: IEEE 2023
• Subjects: Artificial neural networks; Closed loop systems; control barrier functions; Control systems; neural network controller; Nonlinear dynamical systems; safe learning control; Safety; Safety-critical control; Task analysis; Training
• ISSN: 2475-1456; 2475-1456
• DOI: 10.1109/LCSYS.2022.3233322

Learning-based controllers, such as neural network (NN) controllers, can show high empirical performance but lack formal safety guarantees. To address this issue, control barrier functions (CBFs) have been applied as a safety filter to monitor and modify the outputs of learning-based controllers in order to guarantee the safety of the closed-loop system. However, such modification can be myopic with unpredictable long-term effects. In this letter, we propose a safe-by-construction NN controller which employs differentiable CBF-based safety layers and relies on a set-theoretic parameterization. We compare the performance and computational complexity of the proposed controller and an alternative projection-based safe NN controller in learning-based control. Both methods demonstrate improved closed-loop performance over using CBF as a separate safety filter in numerical experiments.